Detecting drug-induced changes in ECG parameters using jacketed telemetry: Effect of different data reduction techniques

Abstract

IntroductionContinuous cardiovascular data is routinely collected during preclinical safety assessment of new medicines. This generates large datasets, which must be summarised to analyse and interpret drug effects. We assessed four methods of data reduction of continuous electrocardiogram (ECG) data and examined the impact on the statistical power of the assay. MethodsContinuous ECG data were collected from a validation study in 6 cynomolgus monkeys using jacketed telemetry. Animals received either vehicle or vehicle followed by ascending doses of moxifloxacin each on a different dosing day. Recordings made for 25h on each dosing day were reduced to discrete time-points using: 1-min average snapshots, 15-min average snapshots, large duration averages (0.5–4h) or super-intervals (3.5–9h averages). ResultsThere was no difference in the ability to detect moxifloxacin-induced QTc prolongation between the 1- and 15-min snapshots and the large duration averages data reduction methods (minimum detectable change in QTca of 20, 17 and 18ms, respectively at 80% power). The super-intervals method detected slightly smaller changes in QTc (15ms), but did not detect a statistically significant increase in QTc after the lowest dose of moxifloxacin, in contrast to the other methods. There were fewer statistically significant differences between dosing days in animals given vehicle when the large duration averages and super-interval reduction techniques were used. DiscussionThere is no marked difference in the power of detection of drug-induced ECG changes in cynomolgus monkeys when using either small duration average or large duration average data reduction techniques. Use of larger duration averages or super-intervals may facilitate data interpretation by reducing the incidence of spurious significant differences that occur by chance between dosing days.